1 Introduction

As the coronavirus spread in Philadelphia this year, travel patterns throughout the city changed. With this shock to human mobility came another to business activity, as many areas saw reliable patrons forced to stay home—especially during the shelter-in-place order which began in March and continued into May. In order to understand the consequences of these changes for business viability, we use mobile phone data to explore the changing nature commercial demand in Philadelphia. This analysis will focus primarily on night life—the restaurants and bars that both provide jobs and support vibrant streets—but we also monitor certain bellwether industries, like those with office work, that doubtless provide a foundation for such night life.

With the goal of understanding the time-space patterns of resident movement in Philadelphia, the following section presents data from SafeGraph, a provider of such records. Note that SafeGraph monitors a representative sample (10%) of the population across the country, so the values shown below are not the true number of visits or journeys, but a slice; that said, we can reduce noise through various aggregations and by tracking trends, which are naturally indexed to population of devices rather than the population as a whole. We use definitions specified in figure 1.1: the number of visitors is the count of devices flowing to a point of interest—be it from a given Census Block Group or total—while a connection is an origin-destination line between a Census Block Group and a point of interest, regardless of its weight. We use both here to determine how many people are moving about the city and from where to where, which calls to how integrated the city is during the pandemic. If one neighborhood is driving business along a corridor or at a veneue, it may obscure the fact that the diversity of clientele—captured in the number of unique neighborhoods sending visitors there—is falling.

1.1 Important definitions

When we link each connection together, we create a network of interactions, weighted by the size of the flow (the number of visits), and in probing the changing structure of this network we can understand the impact of the pandemic on urban life in Philadelphia. This is demonstrated in figure 1.2, where it is evident how much the network thins out as the pandemic erupts.

1.2 Mobility network

Each visit involves a mobile device entering into a point of interest; these include parks and museums, restaurats and bars, or offices and hospitals. In figure 1.3 we map the distribution of these venues and businesses for context. We classify each point of interest by its description, which SafeGraph provides. 1 We can see that most businesses cluster in Center City or nearby but no businesses cluster more than restaurants and bars.

1.3 Business landscape

This analysis comprises different scales, each elaborating on distinct aspects of mobility and activity in Philadelphia. We can look globally, across the city, to explore trends throughout; we can also think locally, dividing the city up into cells or neighborhoods to probe variations within the city. Finally, we can avoid aggregation, looking at individual points or interest or grouping them algorithmically. Below, we attempt to understand the data by engaging them at each scale.

2 Global analysis

In this section we explore trends and relationships manifest most strongly at the global level, across the city. Best described by this focus on the whole over its part are how certain brands and industries are performing, regardless of location, and how certain variables predict changes to activity and mobility in Philadelphia. We see how visitation is changing across time by tracking visits across brands. Figure 2.1 shows that brands associated with necessities (Target and Walmart) saw comparably less of a decline than others, along with fast food restaurants, which one might expect in a time of constrained budgets. The map shows the locations of brands for context.

2.1 Brand performance

Rankings Locations

Figure 2.2, which ranks each brand by the number of visitors it received and animates this change through the pandemic. Dollar stores rise gradually throughout the year, an expected change as residents both need more home goods and need to save money; another important shift is away from non-essential retail towards essential businesses like pharmacies. Starbucks and Wawa occupy top spots for the first several weeks of the year but when the shelter-in-place order sets in, patronage immediately collapses and they are replaced in the ranks essential shops RiteAid and ShopRite.

2.2 Changing fortunes

In figure 2.1 we aggregate by use, grouping by classes like leisure (restaurants and bars) and tourism (museums and theaters). The pandemic had distinct effects on each class, but particularly leisure and other; other includes offices which also explains the steep fall. Interestingly, tourism is recovering while shops and grocers are not, perhaps as many switch to digital commerce.

2.4 Race and travel

The pandemic appears to have flattened an existing relationship between race and mobility: in early days of the pandemic, communities of color were more likely to receive visitors from the rest of the city, a pattern that held for peak months of spread, but this relationship weakens as more predominately white communities gained visitors in July and August. When we plot the same travel patterns against income, we see that wealthy communites are well below their baseline visits, perhaps because many of the restaurant clusters are in relatively affluent areas. While poor communites are more likely to have recovered to baseline but some of the poorest areas are still lagging behind, more in line with wealthy ones.

2.5 Income and travel

The story is similar when we look at inflows, which we document in the appendix. This suggests that poor and minority areas have remained comparable active during the pandemic, which may come with heightened exposure alongside economic stability.

3 Local variation

Perhaps more valuable than probing individual points of interest is aggregating by areal units, which we do in this section. These allow us to see how visits in particular are changing throughout the pandemic in different parts of Philadelphia. Philadelphia has roughly 150 neighborhoods (we use definitions from local firm Azavea) and each responded differently to the pandemic. We explore trends across neighborhoods in figure 3.1; neighborhoods dominated by office work, like the Navy Yard along with Logan Square and Center City, saw precipitous declines in foot traffic, but those with strong amenities and residential communities have recovered. This suggests that demand for food, drink, and shopping may be shifting away from the core. (Note: see the appendix for larger tables.)

3.2 Variation by time and geography

Businesses cluster together and we can explore the strength of this phenomenon by looking at commercial corridors, of which the city has designated roughly 280. Looking at night life in figure 3.3, the largest are Market West and Market East, on either side of city hall, with 1712 and 1263 restaurants and bars respectively, following by Old City at 654 and another in University City with 493: most of the business activity is concentrated in a few locales.

3.3 Commercial corridors

When we plot trends in these clusters over time, it is clear that many of the most successful areas are toward the periphery, perhaps dormitory communities supported by remote work, and several of the least successful are situated in the core. Notably among the worst performers are the two central corridors, which depend on office work, and the Sports Complex, which saw sports leagues take measures of protect players and ban fans early on—and many of these restrictions are still in place. Peripheral plazas like Oxford and Levick, home to a supermarket, and City and Haverford are among the best.

3.5 Explanatory variables

4 Points of Interest

This section looks at individual points of interest, how they perform over time and whether or not we can identify certain bellwether businesses within the city. These cases can provide further insight into how the pandemic is changing mobility. We start by looking at the network of connections across the city. Drawing a line between each origin (neighborhood) and destination (point of interest), there is a dense web—a nearly saturated graph where all neighborhoods send visitors to all other corners of the city. This web becomes sparser as the pandemic came to the fore and there during the late summer there were fewer links than during the late winter.

4.1 Aggregate mobility

As we saw above, the data show that big box stores like Target and Walmart appear to have weathered the pandemic well, but the shift to remote work should also appear in the data. We can look at visits to the Comcast Center and the Plaza below it; visits in April and May, as the coronavirus took hold in the city, fell substantially.

4.2 Business in focus

Yet with offices vacant, parks should have swelled with visitors. We see mixed evidence of this in the data. Philadelphia has four central squares—Rittenhouse, Washington, Logan, and Franklin—which provide important community amenity; all saw fewer visits in April and May than later in the summer, suggesting winter patterns continued even as the weather improved. As a signal for tourism, we can look at Reading Terminal Market; vendors between its walls saw marked declines in visits beginning in April.

4.3 Tourism in focus

Appendix

A.1

Best Worst

A.2

  1. If that description contains “restuarant” or “bar”, we call that leisure. Anything educational, from tutoring to public, private or charter schools to tertiary education, we call that education. Tourism includes museums and parks.↩